Você está na página 1de 8

Newtons Radii, Maupertuis Arc Length,

and Voltaires Giant


Andrew J. Simoson
Andrew Simoson (ajsimoso@king.edu) is Professor of
Mathematics at King College in Tennessee. His recent
book, Voltaires Riddle (MAA, 2010), describes the
remarkable agreement of contemporary satellite data with
the arc length measurements of 18th-century French
mathematicians. Here he is alongside the Maupertuis
monument in Tornio, Finland, where he gave an invited
address this past summerat the arctic circle near the
center of the French expeditions network of nine triangles
used to estimate that arc length.
When Isaac Newton proposed a theory of gravitation, he concluded that our rotat-
ing earth must be attened at its poles, so that the polar radius would be
229
230
that of
the equatorial radius, and additionally estimated this difference in radii, r, as 17.1
miles [4, p. 824]. Unconvinced, continental mathematicians, by and large, long main-
tained that the earth bulged at its poles. As an effort to resolve this, by then forty-plus-
year, stalemate, Pierre-Louis Moreau de Maupertuis successfully lobbied the French
Academy of Sciences to launch expeditions both to the far north and to the equator to
measure the arc length s along lines of longitude on the earths surface. By the summer
of 1737, Maupertuiss team had returned from Lapland with the arc length along one
degree of longitude near the arctic circle, denoted s, as 69.52 miles (57,395 toises)
[2, p. 227]. Not long thereafter, Voltaire featured Maupertuis team of mathematicians
in a scientic romance, Microm egas, wherein Microm egas, an alien giant 23 miles
tall, questions the team about how they measured the quantities that they did and about
measurement in general.
What relationships are there among these three numbers: r, s, and Microm egass
height? The answer involves a little calculus. The basic problem is to nd the major
and minor axes of an ellipse given some information about its arc length. We give a
simplied version of Newtons argument, then show how an elliptical prole model
for the earths shape together with an estimate of s determines r. We close by
speculating why Voltaires ctional giant is precisely 23 miles tall.
A modied Newtonian formula
We arrive at an estimate for r closer than Newtons to its true value by simplifying
his reasoning. Let R be the earths polar radius, its equatorial radius, and its ro-
tation rate (2 radians per day). Let g be the acceleration due to gravity on a particle
at the north pole, and P be a particle on the equator. What is the acceleration on P
due to the earths mass alone? In a very simple model, which we call the homogeneous
model, the acceleration on a particle at distance r from a spherical earths center is
doi:10.4169/college.math.j.42.3.183
VOL. 42, NO. 3, MAY 2011 THE COLLEGE MATHEMATICS JOURNAL 183
linear, namely, (r) = gr/R. This model was in vogue throughout the eighteenth cen-
tury, although some intractable models were also considered such as planetary densi-
ties resembling a badly-stirred pudding or pathologies wherein gravity changes with
longitude [2, p. 254]. The outward acceleration on P due to centrifugal force is
2
.
Thus the total acceleration on P is g/R
2
. Let us now assume, navely, that
on a planet consisting entirely of water (hence homogeneous), surface gravity is con-
stant, intuition suggesting that water might ow so as to reach a surface equilibrium
with respect to gravity, being a combination of acceleration due to mass and rotation.
Solving
g
R

2
= g,
for gives
= R

1
1

2
R
g

. (1)
Supposing R is between 4000 100 miles and g = 32 ft/sec
2
, then
2
R/g is much
smaller than 1; in particular, if R < 4100 miles, then
2
R/g < 0.0036, a unitless
quantity. Using a geometric series in (1) gives
R

1 +

2
R
g

,
so that
r = R

2
R
2
g
. (2)
With R between 4000 100 miles, Newtons modied formula (2) gives r
between 13.3 and 14.7 miles. The actual value is 13.3 miles according to satel-
lite measurements. Furthermore, if we use Newtons values of g and R, namely
g 31.89 ft/sec
2
, and R 3914.6 miles [4, pp. 822824], then (2) gives 13.4 miles.
Caveat
Newton, however, would have been dissatised with this result. Lacking a way to
check its accuracy against the true value of r, he would object that in deriving (2),
we used a radially symmetric acceleration gr/R, whereas with an elliptical prole,
acceleration on particles within a homogeneous earth is no longer radially symmetric.
To resolve this dilemma, Newton proceeded in a more careful manner, that included
the calculation of two triple integrals of the inverse square law once with respect to
the north pole and and again with respect to a point on the equator over a homoge-
neous ellipsoidalthough Newton called it making the computation [4, p. 823], to
conclude that r is about 17.1 miles.
Although the model of a homogeneous earth gives good results for r, the model is
fundamentally awed because the earths density is far from constant. Of course, prac-
tical implementation of variable density models had to await Lord Henry Cavendishs
1798 experimental determination of the value of G, the universal gravitation constant.
Prior to his results, only the product GM was known, where M is the mass of the earth.
184 THE MATHEMATICAL ASSOCIATION OF AMERICA
With much work, twentieth century geophysicists using seismograph analysis have
determined that, whereas the acceleration function at the earths surface is approx-
imately g = 32 ft/sec
2
, (r) attains an extreme of about 35.64 feet/sec
2
[3, p. 155]
near the junction marked M on Figure 1 between the liquid outer core and the mantle
at r 2190 miles, where r is distance from the center. The junction marked C on
Figure 1 is between the inner iron core and the liquid outer core. Using this data
for a second simple model, which we call the variable density model, how does (2)
change? Lets assume an afne relation between (r) and r near r = R, that is,
(r) =

(R)(r R) + g, where

(r) is the derivative of (r), and also take, as a


crude approximation for

(R), the slope m, where m = (g 35.64 ft/sec


2
)/(R
2190 miles) 0.00000038 sec
2
. Thus the difference between acceleration due to
mass and due to rotation is m(r R) + g r
2
.
h
o
m
o
g
e
n
e
o
u
s

m
o
d
e
l
outer core mantle outer space
Distance r from earths center
A
c
c
e
l
e
r
a
t
i
o
n

(
r
)

i
n

f
t
/
s
e
c
2
32
0 C M R
r
Figure 1. Acceleration versus distance r from earths center, adapted from [3, p. 156].
Rather than letting r = in this last expression and setting it equal to g, as we
did to generate (2), we set it equal to the value of overall acceleration at the equator.
That is, since the overall acceleration at the equator is less than at the poles by about
= 0.17 ft/sec
2
[3, p. 49], we have
m( R) + g
2
= g .
Solving for and then subtracting R gives
r = R =

2
R
m
2
28.6 miles, (3)
not nearly so good a result as with the simpler model. Of course, the difculty with
(3) is that our choice for m is poor. Since so much physical data needs to be known to
calculate r with this model, a better use for it is to estimate the left hand derivative of
(r) at r = R. That is, knowing r and , we estimate that the left hand derivative of
(r), namely m, is probably about 2.1 times the value we had chosen initially, a matter
whose verication we leave to geophysicists. What the variable density model really
demonstrates is that both Newtons approach and our simplication were extremely
lucky to obtain such good results using a homogeneous model.
VOL. 42, NO. 3, MAY 2011 THE COLLEGE MATHEMATICS JOURNAL 185
How s gives r
Let us assume that the earth is an ellipsoid, and that its prole is given by the parametric
equations of an ellipse:
(x, y) = ( cos , R sin ),
where 0 2. By latitude at a point P on this ellipse, we mean the angle
between a normal to the ellipse at P and a line through P parallel to the x-axis. How
does relate to ?
Figure 2 shows the prole of a planet where = 2 and R = 1. The slope of the
tangent line to the parametrized curve at is given by
dy
dx
=
dy/d
dx/d
=
R cos
sin
=
R

cot ,
which means that the latitude corresponding to parameter is
() = tan
1

R
tan

,
or, equivalently, the parameter corresponding to latitude is
() = tan
1

tan

.
Meanwhile the polar angle corresponding to parameter is
() = tan
1

y
x

= tan
1

tan

.
For the ellipse of Figure 2, if = 66.5

N, the latitude of the arctic circle, then


s radian measure is 1.16064 radians, which corresponds to the parameter value
(1.16064) 0.855019, so that (0.855019) 0.521805 radians, or about 29.9

.
= 29.9
= 66.5
corresponding to = 0.885019
x
y
Figure 2. An extreme model of the earths prole.
186 THE MATHEMATICAL ASSOCIATION OF AMERICA
To use Maupertuiss value of s = 69.52 miles to determine the earths shape, we
take
1
and
2
as the radian measures of 66

and 67

. Then
1
= (
1
) and
2
= (
2
).
The arc length of one degree along the meridian at the arctic circle is therefore
Q
a
(, R) =

2
sin
2
+ R
2
cos
2
d, (4)
(where a represents the arctic circle). To nd and R which make s 69.52 miles,
we plot in Figure 3, all pairs (, R) for which equatorial and polar radii are within
50 miles of 4000. The region > R contains Newtonian models. Since the Cassinis
(father and son, astronomers in Paris) championed an earth shaped like a lemon rather
than the mandarin orange of Newton, the region < R contains Cassinian models.
The thick black path through this plot characterizes those (, R) values for which
|Q
a
69.52| 0.1. The white line through this black path is R = 3.03 8086.36.
We see that r ranges from45 miles to 22 miles. In particular, if R 3950.8 miles,
earths actual polar radius as determined from satellite data, then 3972.3, giving
r 21.5 miles, a value propitiously close to Microm egass height.
3960 3980 4000 4020 4040
3960
3980
4000
4020
4040
Newtonian models
Cassinian models
P
o
l
a
r

r
a
d
i
u
s

R

i
n

m
i
l
e
s
Equatorial radius in miles
Figure 3. Possible equatorial and polar radii.
A second data point
With a second arc length data point, we can be much more precise about the earths
shape. A natural second data point is the measurement obtained by the French equato-
rial team sent to what is now Ecuador in 1735. Although that expedition began a year
before Maupertuis, they arrived at the measurement of 68.76 miles (56,768 toises) for
one degree of arc at the equator in 1744 [2, p. 227]. Now let
Q
e
(, R) = 2


3
0

2
sin
2
+ R
2
cos
2
d, (5)
where
3
is the parameter value associated with the latitude of 0.5

N (and where e
represents the equator). How can we nd the and R values that best reect the two
VOL. 42, NO. 3, MAY 2011 THE COLLEGE MATHEMATICS JOURNAL 187
given arc length values? We use least squares, and minimize
M=

(Q
a
69.52)
2
+(Q
e
68.76)
2
,
where Q
a
and Q
e
are from (4) and (5). Figure 4 is a contour plot of M. The solid black
oval represents all values (, R) for which Mis no larger than 0.01 miles. Hence the
two geodesic expeditions launched by Louis XV give 3974.2 and R 3956.9
miles, for a difference of r 17.3 miles, uncannily close to Newtons original esti-
mate of 17.1 miles.
3973.6 3973.8 3974.0 3974.2 3974.4 3974.6 3974.8 3975.0
3956.0
3956.2
3956.4
3956.6
3956.8
3957.0
3957.2
3957.4
Equatorial radius in miles
P
o
l
a
r

r
a
d
i
u
s

R

i
n

m
i
l
e
s
Figure 4. The polar and the equatorial expeditions and R.
We say uncanny since in later years it was determined that Maupertuis arctic team
overestimated by about 0.25 miles. With this new arc length value of 69.27 miles and
the equatorial teams old value of 68.76 miles, we get 3962.8 miles, R 3951.3
miles, so that r 11.5 miles, not too far aeld from the satellite measurements of
3964.1 miles, R 3950.8 miles, and r 13.3 miles.
Suppose instead we take the two arc lengths in Newtons Principia. There one de-
gree centered halfway between London and York at latitude 0.9203 radians is given
as 69.41 miles (57,300 toises) and another degree centered halfway between Corbeil
(just south of Paris) and Amiens at latitude 0.8593 radians is 69.12 miles (57,060
toises) [4, p. 822] and [6, pp. 6777]. Then the (, R) values that minimize the square
root of the sum of the squares of the error is (3987, 3895), giving r 92 miles,
which means that these two measurements are really inconsistent with the model of
the earths shape being almost a sphere! Of course, these measurements were taken
years apart using different instruments; in 1637, Richard Norwood used a chain of
length 99 feet (6 rods) for the former measurement, while in 1670, Jean Picard used
a triangularization scheme for the latter. So perhaps this result is not quite so unex-
pected. Furthermore, Newton used these two measurements, not to compute r, but
to cite data supporting the mandarin orange model, that arc length per degree increased
with latitude.
188 THE MATHEMATICAL ASSOCIATION OF AMERICA
A curious footnote
What about Voltaires giant?
Voltaire was an acionado of Newton. In fact, Voltaires popular 1738 book, Ele-
ments of Newtons Philosophy, is an introduction to Newtons theories for the reading
public. He recruited Maupertuis and others to help him write it. When Maupertuis
returned with data supporting Newtons model of the earth, Voltaire called him Flat-
tener of the Earth and of the Cassinis, too [7, p. 161].
In Microm egas, a giant intercepts the French mathematicians on their return to
Paris, 23 miles tall, or, as Voltaire also phrases it, 24,000 times the height of a ve
foot man. What was Voltaire thinking?
Voltaire enjoyed toying with the speculations of savants. For example, Kepler, in
a letter to Galileo congratulating him on the discovery of Jupiters moons, [1, p. 93]
wrote, I long for a telescope to anticipate you in discovering two moons around Mars.
Voltaire has Microm egas on his way to earth espy two Martian moons, moons only
later discovered in 1877 by Asaph Hall.
In like fashion, Voltaire may have played with a wild speculation of Jacques Cassini.
In 1717, Cassini suggested that Sirius, the brightest star in the night sky, had radius
equal to 100 radii of the sun. Todays astronomers put its radius as only 1.75 that of
the sun. Now Microm egas just happens to be from Sirius. And how big did Voltaire
make his ctional Sirius? It turns out its radius is a natural one with respect to the earth,
and one that surpasses Cassinis guess. At the opening of his story, Voltaire writes and
solves an exercise [5, p. 8]:
Some algebraistspeople ever useful to the publicat once took their pens and
since Monsieur Microm egas is 120,000 feet from head to toe while we earthlings
are barely ve feet and since earth is 9,000 leagues around, estimated his world
to be 216 million leagues around.
Now 216 million leagues is the length of the earths orbit about the sun: Voltaires
league is about 2.8 miles, so that 216 million leagues divided by 2 is 96 million
miles, approximately one astronomical unit, or more than double Cassinis guess.
There may be more. As we have seen, behind Microm egas is the quest for r.
Microm egass height being on the same order of magnitude as 17.1 miles makes the
scene of the French mathematicians return from the arctic circlewhere they had
painstakingly measured s miles of territory, biting-y-infested during the summer
and bitterly-cold during the winter, so as ultimately to quantify rmeeting a crea-
ture whose very height is approximately the answer to their question quaintly ironic.
Just as Voltaire inated Cassinis guess about the size of Sirius, he may have chosen
to inate Newtons guess to 23 miles, nding thereby a natural number both represen-
tative of the ratio of an astronomical unit to the earths radius and of the same order of
magnitude as 17.1.
This chain of suppositions may stretch the ingenuity even of a Voltaire. Still the
giants incredible height ts the nature of the story. As Newton himself said, I stand
on the shoulders of giants. So too, Microm egas afrmed with explicit 23-mile tall
imagerylong before satellite measurements conrmed itthat the giant Newtons
guess for r was very, very good.
Summary. Given two arc length measurements along the perimeter of an ellipseone taken
near the long diameter, the other taken anywhere elsehow do you nd the lengths of major
and minor axes? This was a problem of great interest from the time of Newtons Principia
until the mid-eighteenth century when France launched twin geodesic expeditionsone to
VOL. 42, NO. 3, MAY 2011 THE COLLEGE MATHEMATICS JOURNAL 189
the equator near Quito, the other to the Finnish arctic led by Maupertuisso as to determine
whether the earth was lemon-shaped, as the French Academy long contended, or like a man-
darin orange as Newton promised. We give a simplied version of Newtons argument, and
show how an elliptical prole model for the earths shape together with an arc length measure-
ment determines the amount of attening of the earth at the poles. We conclude by speculating
why Voltaire made his giant Microm egas exactly 23 miles tall.
References
1. S. H. Gould, Gulliver and the moons of Mars, J. Hist. of Ideas 6 (1945) 91101. doi:10.2307/2707059
2. Michael Rand Hoare, The Quest for the True Figure of the Earth, Ashgate Press, Farnham UK, 2005.
3. William Lowrie, Fundamentals of Geophysics, Cambridge University Press, Cambridge UK, 1997.
4. Isaac Newton, The Principia, Mathematical Principles of Natural Philosophy, trans. by I. Bernard Cohen and
Anne Whitman, University of California Press, Berkeley CA, 1999.
5. Andrew Simoson, Voltaires Riddle: Microm egas and the measure of all things, Mathematical Association of
America, Washington DC, 2010.
6. James R. Smith, From Plane to Spheroid: determining the gure of the earth from 3000 BC to the 18th Century
Lapland and Peruvian survey Expeditions, Landmark Press, Boulder CO, 1986.
7. Mary Terrall, The Man who Flattened the Earth: Maupertuis and the Sciences in the Enlightenment, University
of Chicago Press, Chicago, 2002.
Every 4:00 A.M. Substance is the rst of six cats to arise and
go rummaging through the wastebasket.
He needs something, anything.
It doesnt have to be food.
As long as he can chase it around.
As long as it scratches along the oor.
Devin says hes frightened.
Frightened of everything.
But I say Substance is looking for something to prove.
So give him denitions.
Notation.
Axioms.
Yes. Substance needs some axioms
to scratch along the oor.
from Crossing the Equal Sign by Marion Deutsche Cohen [p. 30]
reviewed in this issue on page 241
190 THE MATHEMATICAL ASSOCIATION OF AMERICA

Você também pode gostar